Prosodic resolution of syntactic ambiguity

This paper deals with the use of prosodic information in disambiguating syntactic structure, i.e. in arriving at the correct interpretation of a syntactically ambiguous sentence. We will start out by examining closer the linguistic phenomenon of prosody, its acoustic and functional properties. It is pretty clear that prosody plays an important role in human speech understanding, however, the exact nature of this role has been heavily investigated within the (psycho)linguistic literature, and we will accordingly review some of these studies and attempt to arrive at an approximation regarding the main findings there. Whether knowledge about prosody also can be utilized in automatic speech understanding is a relevant question in the present context. We will, as in Nöth et al. (2000), differentiate between automatic speech recognition (ASR) and speech understanding (ASU), where the former refers to the process of converting an acoustic signal into a string of words and the latter the application of further linguistic processing, such as syntactic parsing, to this string of words. For the purpose of this paper, the main focus will be on automatic speech understanding. This represents a difficult task, which is complicated further by syntactic ambiguity, as many other systems of natural language processing. We will therefore review some of the work done on incorporating prosodic information into syntactic parsing and by doing so attempting to provide a better understanding of the problems faced and the practical viability of such an incorporation within actual systems of speech understanding.